Beware of ICD-10 shortcuts: The case against crosswalking
Monday, March 02, 2015
ICD-10 is inherently more sophisticated and specific than ICD-9, but that doesn't necessarily mean it's more complicated. So, why are healthcare professionals pulling their hair out over the mandatory transition to these new codes? Because learning ICD-10 is like learning a new language.
If we were going into this with a clean slate — like a newborn babe — perhaps it wouldn't be so tough to learn the language. Unfortunately, the U.S. healthcare industry has relied on ICD-9 codes for more than 30 years. Can you imagine learning to speak another language fluently after 30 years of speaking only one tongue?
Obviously, we've got a lot of relearning to do. That's what's complicated — the re-education. And healthcare providers are becoming increasingly aware of the time and monetary investment re-education requires.
It's an investment that has to happen, though, because "insufficient preparation in implementing the transition to ICD-10 will result in inaccurate coding, escalating queries, delayed billing, and denials of reimbursements," ICDLogic explains in this whitepaper.
So, re-education — and the time, money and effort that goes with it — is absolutely necessary, and it needs to happen quickly. After all, the Oct. 1, 2015, transition date is looming, and the next seven-or-so months are going to fly by in the blink of an eye.
This is when crosswalking — either in the form of General Equivalence Mappings (GEMs) or other automated tools — becomes most appealing. However, as we learned with functional limitation reporting, crosswalks may look like the easiest path to preparation, but they — like language translation dictionaries — could end up making you look like a fool.
Let's examine why.
GEMs are bad, mmmkay
Okay, that's an exaggeration. Created by CMS and the National Center for Health Statistics (NCHS), GEMs have an actual purpose — to convert large batches of data, which, as EHR Intelligence explains, is useful for:
- Linking data in clinical studies that span Oct. 1, 2015
- Analyzing data collected before and after the transition
- Translating an application's code lists or tables when it only has one code set as its source
- Calculating equivalent reimbursements across code sets and refining reimbursement and quality applications
Conversely, healthcare providers should not use GEMs for translating small batches of data, like those that exist within patient documentation or claims. Furthermore, CMS and NCHS developed GEMs for use during the transition period — not after. Thus, these organizations will not update or maintain GEMs after Oct. 1, 2015.
Translation: GEMs are not your handy-dandy, go-to ICD-10 dictionary or crosswalking BFF.
Now, let's say you want to use a GEM the right way — for a large data conversion. That's fine, but the job doesn't end there.
Once you use the GEM to convert ICD-9 to ICD-10 codes, you then need to manually verify the accuracy of all the conversions. Why? Because there are so many more ICD-10 codes than ICD-9 codes (about 68,000 compared to 13,000).
In fact, in some cases there are dozens — or even hundreds — of possible ICD-10 equivalents for a single ICD-9 code. This means that humans must play a role in the translation process.
As EHR Intelligence explains, "Because accurate documentation requires the selection of the most specific applicable code available, human eyes are needed to ensure that an ICD-10 code actually matches the patient's condition. If a vague or inaccurate code is selected, payers may not reimburse for the accompanying services."
GEMs can translate large batches of data, but because they do not produce a reliable one-to-one translation, humans must still verify the accuracy of that translation. You shouldn't use GEMs for case-by-case crosswalking, and GEMs are not a substitute for learning how to select ICD-10 codes.
As Cognizant explains in this whitepaper, "a crosswalk-based approach ... is fraught with complexity and added expense and ... the need for manual review of data and claims."
Buyer beware: Automated crosswalking software
To make matters worse, the GEMs created by CMS and NCHS aren't the only automated crosswalking tools. There are no industrywide standards when it comes to GEMs; any entity can create and distribute them.
As ICDLogic states in this whitepaper, it's crosswalking chaos out there: "Today there are many GEMs available from a variety of sources, EHR vendors and payers. No one really knows whether they are consistent or what the individual organizing principles and assumptions are that each developer used." Yikes.
Furthermore, when it comes to translating a language, context is everything. The same holds true for translating diagnosis codes. Patient documentation and the practitioner's own clinical opinions factor into the diagnosis.
Thus, if you're relying on a software that proclaims it can "handle" ICD-10 for you and produce "one-to-one matches" without any effort on your part, you should exercise extreme caution, because you have no idea how this vendor arrived at such one-to-one translations.
Plus, as I already emphasized, it's impossible for any practitioner to arrive at accurate one-to-one translations without taking into account the patient in front of him or her. When it comes to converting ICD-9 to ICD-10, there are "445 instances where a single ICD-9 code can map to more than 50 ICD-10 codes," and there are "210 instances where a single ICD-9 can map to more than 100 ICD-10 codes," says ICDLogic.
In that same whitepaper, ICDLogic cites a UnitedHealthcare report indicating that "only about 5 percent of all codes will map accurately 1:1." So, if any software — regardless of specialty — asserts that a high percentage of codes map one-to-one, be especially wary.
Remember, payment is on the line. Do you really want to take any chances?
Moral of the story
There is not a single automated software or tool that can take a particular ICD-9 code and spit out an exact ICD-10 match. Sure, there are tools that can get you close, but you, as the practitioner, will need to put some effort into landing on the final diagnosis and its corresponding ICD-10 codes — no ifs, ands or buts about it.
OK, so GEMs and other automated one-to-one crosswalking tools aren't the way to go. What, then, is the right path? You have a few options:
Start with a conversion tool, then do some legwork
- Enter the ICD-9 code you'd normally use into a conversion tool, and use the resulting ICD-10 code as your starting point.
- Find that ICD-10 code within the Tabular List and determine whether there are codes with greater levels of specificity available.
- Pay attention to chapter and category headings to ensure you factor any additional coding instructions into your selection.
- If possible, code for the actual condition (e.g., patellar tendinitis) instead of the result (e.g., knee pain).
- If required, use pertinent external cause codes, which you can find in Chapter 20 of the Tabular List.
Find the right EMR
You're the expert; you have the patient in front of you, and you know the most accurate diagnosis and treatment plan. So, if you already have an EMR or are investigating EMR, find out how the company plans to handle ICD-10.
You should choose an EMR solution that offers an ICD-10 tool that puts you in the driver's seat. The tool should integrate with the documentation portion, so it helps you select the correct, most specific diagnosis based on the patient note.
Ultimately, you should select an ICD-10 tool that gives you the freedom to express patient diagnoses more accurately and succinctly using the new coding language. In short, do not go with an unreliable translation dictionary or one-to-one crosswalker.
- 17 of the most specific, bizarre ICD-10 codes
- BSN or ADN? Nursing at a crossroads
- Nurses rally in DC to address staffing issues with Congress
- Why telemedicine is the future of healthcare
- Big winners in California’s new healthcare plan: Households and small businesses
- You cannot lead until you have their trust
- Nurses to rally in DC again to promote safe staffing
- Nurse coaching on the rise
- Employers aren’t worried about unethical AI, but maybe they should be
- 3 words to describe your association
- The blessing and burden of caregiving
- Global goods giant commits to drastically reducing plastic packaging by 2025
- Chiricahua National Monument: The land of standing rocks
See your work in future editions
Your content, Your Expertise,
Your Industry Needs YOUR Expert Voice & We've got the platform you needFind Out How